I don't know the current drive capabilities of the Broadcom SoC, but let's assume it can perfectly drive a dipole antenna presenting around 72 ohms. The supply is 3.3V, I think. 3.3^2/72 = 0.15125W. In reality I doubt it's anywhere close to that, let's assume 50mW instead (it could be lower, I just don't know). According to a legit-looking field strength calculator I found online this works out to just over 0.5V/m at 3m! That sounds a bit high. Various factors would conspire to making even this much field strength unlikely, but it'd still exceed the FCC limits easily.

The Pi's output is also a reasonably decent square wave, as pointed out elsewhere. So there are some harmonics that are, themselves, above the part 15 limits.

In other words, don't do this, unless you have a spectrum analyzer, attenuator (don't forget to block DC). Actually, someone needs to do this with the right equipment just to find out how well it really performs.